checkpoint_0 of transfer learning dosen't sound like the pre-trained model

NVIDIA / tacotron2

Tacotron 2 - PyTorch implementation with faster-than-realtime inference

BSD 3-Clause "New" or "Revised" License

5.12k stars 1.39k forks source link

checkpoint_0 of transfer learning dosen't sound like the pre-trained model #562

Open jhonmaker-gonsu opened 2 years ago

jhonmaker-gonsu commented 2 years ago

I think first checkpoint for transfer learning should sound like pre-trained model,But the model sound gets noisy and pronouciation is a bit fuzzy. both sampling rate is 22KHz Can anybody help ? Thanks.